Hierarchical analysis of medical images for identifying and assessing lymph nodes

Abstract

Systems and methods for identifying and assessing lymph nodes are provided. Medical image data (e.g., one or more computed tomography images) of a patient is received and anatomical landmarks in the medical image data are detected. Anatomical objects are segmented from the medical image data based on the one or more detected anatomical landmarks. Lymph nodes are identified in the medical image data based on the one or more detected anatomical landmarks and the one or more segmented anatomical objects. The identified lymph nodes may be assessed by segmenting the identified lymph nodes from the medical image data and quantifying the segmented lymph nodes. The identified lymph nodes and/or the assessment of the identified lymph nodes are output.

The invention claimed is: 
     
       1. A method comprising:
 receiving medical image data of a patient; 
 detecting one or more anatomical landmarks in the medical image data; 
 segmenting one or more anatomical objects from the medical image data based on the one or more detected anatomical landmarks; 
 identifying one or more lymph nodes in the medical image data based on the one or more detected anatomical landmarks and the one or more segmented anatomical objects using a machine learning network, the machine learning network 1) receiving as input the medical image data having voxels labelled as corresponding to the one or more anatomical objects in accordance with the segmenting, 2) evaluating only voxels of the medical image data that are not labelled as corresponding to the one or more anatomical objects, and 3) generating as output a probability map indicating a probability score of voxels belonging to the one or more lymph nodes to identify the one or more lymph nodes in the voxels that are not labelled as corresponding to the one or more anatomical objects; and 
 outputting the one or more identified lymph nodes. 
 
     
     
       2. The method of  claim 1 , wherein the machine learning network is a U-Net trained to map intensities of voxels of the medical image data as a Gaussian volume. 
     
     
       3. The method of  claim 1 , further comprising:
 assessing the one or more identified lymph nodes. 
 
     
     
       4. The method of  claim 3 , wherein assessing the one or more identified lymph nodes comprises:
 segmenting the one or more identified lymph nodes from the medical image data; and 
 quantifying the one or more segmented lymph nodes. 
 
     
     
       5. The method of  claim 1 , wherein detecting one or more anatomical landmarks in the medical image data comprises:
 detecting the one or more anatomical landmarks in the medical image data using an agent trained with deep reinforcement learning. 
 
     
     
       6. The method of  claim 1 , wherein segmenting one or more anatomical objects from the medical image data based on the one or more detected anatomical landmarks comprises:
 segmenting the one or more anatomical objects from the medical image data using an adversarial deep image-to-image network. 
 
     
     
       7. The method of  claim 1 , wherein the medical image data comprises one or more computed tomography images. 
     
     
       8. An apparatus, comprising:
 means for receiving medical image data of a patient; 
 means for detecting one or more anatomical landmarks in the medical image data; 
 means for segmenting one or more anatomical objects from the medical image data based on the one or more detected anatomical landmarks; 
 means for identifying one or more lymph nodes in the medical image data based on the one or more detected anatomical landmarks and the one or more segmented anatomical objects using a machine learning network, the machine learning network 1) receiving as input the medical image data having voxels labelled as corresponding to the one or more anatomical objects in accordance with the segmenting, 2) evaluating only voxels of the medical image data that are not labelled as corresponding to the one or more anatomical objects, and 3) generating as output a probability map indicating a probability score of voxels belonging to the one or more lymph nodes to identify the one or more lymph nodes in the voxels that are not labelled as corresponding to the one or more anatomical objects; and 
 means for outputting the one or more identified lymph nodes. 
 
     
     
       9. The apparatus of  claim 8 , wherein the machine learning network is a U-Net trained to map intensities of voxels of the medical image data as a Gaussian volume. 
     
     
       10. The apparatus of  claim 8 , further comprising:
 means for assessing the one or more identified lymph nodes. 
 
     
     
       11. The apparatus of  claim 10 , wherein the means for assessing the one or more identified lymph nodes comprises:
 means for segmenting the one or more identified lymph nodes from the medical image data; and 
 means for quantifying the one or more segmented lymph nodes. 
 
     
     
       12. A non-transitory computer readable medium storing computer program instructions, the computer program instructions when executed by a processor cause the processor to perform operations comprising:
 receiving medical image data of a patient; 
 detecting one or more anatomical landmarks in the medical image data; 
 segmenting one or more anatomical objects from the medical image data based on the one or more detected anatomical landmarks; 
 identifying one or more lymph nodes in the medical image data based on the one or more detected anatomical landmarks and the one or more segmented anatomical objects using a machine learning network, the machine learning network 1) receiving as input the medical image data having voxels labelled as corresponding to the one or more anatomical objects in accordance with the segmenting, 2) evaluating only voxels of the medical image data that are not labelled as corresponding to the one or more anatomical objects, and 3) generating as output a probability map indicating a probability score of voxels belonging to the one or more lymph nodes to identify the one or more lymph nodes in the voxels that are not labelled as corresponding to the one or more anatomical objects; and 
 outputting the one or more identified lymph nodes. 
 
     
     
       13. The non-transitory computer readable medium of  claim 12 , wherein detecting one or more anatomical landmarks in the medical image data comprises:
 detecting the one or more anatomical landmarks in the medical image data using an agent trained with deep reinforcement learning. 
 
     
     
       14. The non-transitory computer readable medium of  claim 12 , wherein segmenting one or more anatomical objects from the medical image data based on the one or more detected anatomical landmarks comprises: segmenting the one or more anatomical objects from the medical image data using an adversarial deep image-to-image network.